Refactor 2 (#36)

Improve the filtering of data based on common support in propensity
scores. The most important changes include importing utility functions,
refactoring the `filter_common_support` function, and adding new utility
functions for better code modularity.

### Addition of new utility functions:

*
[`CausalEstimate/filter/propensity.py`](diffhunk://#diff-73e7b9da6dfdab6a8a1711b59df52d4bffac8ad5d5a5db07f7ddc00b69acfa89L23-R50):
Added `get_common_support_range` function to calculate the common
support range for propensity scores, improving the readability and
maintainability of the code.
*
[`CausalEstimate/utils/utils.py`](diffhunk://#diff-47ddc6f8fde028e7680ca9bc39bbe60f9539d833069d20a5d4ce9fac195f2396R18-R23):
Added `filter_column` function to filter a DataFrame based on a
specified column range, enhancing the modularity of the code.

CausalEstimate/filter/propensity.py

@@ -1,4 +1,5 @@
 import pandas as pd
+from CausalEstimate.utils.utils import get_treated_ps, get_untreated_ps, filter_column
 
 
 def filter_common_support(
@@ -20,19 +21,36 @@ def filter_common_support(
     Returns:
     DataFrame after removing individuals without common support.
     """
-    # Split the dataframe into treated and control groups
-    treated = df[df[treatment_col] == 1]
-    control = df[df[treatment_col] == 0]
-    # Get the range of propensity scores for treated and control groups
-    min_ps_treated, max_ps_treated = treated[ps_col].quantile(
-        [threshold, 1 - threshold]
+    common_min, common_max = get_common_support_range(
+        df, treatment_col, ps_col, threshold
     )
-    min_ps_control, max_ps_control = control[ps_col].quantile(
+    filtered_df = filter_column(df, ps_col, common_min, common_max)
+    return filtered_df
+
+
+def get_common_support_range(
+    df: pd.DataFrame, treatment_col: str, ps_col: str, threshold: float = 0.05
+) -> tuple[float, float]:
+    """
+    Calculate the common support range for propensity scores.
+
+    Parameters:
+    -----------
+    df : Input DataFrame with treatment and propensity score columns.
+    treatment_col : Name of the treatment status column.
+    ps_col : Name of the propensity score column.
+    threshold : Quantile threshold for trimming score distribution tails. Default is 0.05.
+
+    Returns:
+    --------
+    Lower and upper bounds of the common support range.
+    """
+    min_ps_treated, max_ps_treated = get_treated_ps(df, treatment_col, ps_col).quantile(
         [threshold, 1 - threshold]
     )
-    # Define the common support range
+    min_ps_control, max_ps_control = get_untreated_ps(
+        df, treatment_col, ps_col
+    ).quantile([threshold, 1 - threshold])
     common_min = max(min_ps_treated, min_ps_control)
     common_max = min(max_ps_treated, max_ps_control)
-    # Filter individuals to keep only those within the common support range
-    filtered_df = df[(df[ps_col] >= common_min) & (df[ps_col] <= common_max)]
-    return filtered_df
+    return common_min, common_max

CausalEstimate/utils/utils.py

@@ -15,3 +15,10 @@ def get_treated(df: pd.DataFrame, treatment_col: str) -> pd.DataFrame:
 
 def get_untreated(df: pd.DataFrame, treatment_col: str) -> pd.DataFrame:
     return df[df[treatment_col] == 0]
+
+
+def filter_column(df: pd.DataFrame, col: str, min: float, max: float) -> pd.DataFrame:
+    """
+    Filters a DataFrame to keep only rows where a specified column is within a given range.
+    """
+    return df[(df[col] >= min) & (df[col] <= max)]